Dual apodization with cross-correlation combined with robust Capon beamformer applied to ultrasound passive cavitation mapping

Purpose: Passive acoustic mapping (PAM) has received increasing attention in recent years and has an extremely widespread application prospect in real-time monitoring of ultrasound treatment. When using a diagnostic ultrasound transducer, such as a linear-array transducer, the initially used time exposure acoustics (TEA) algorithm will produce high-level artifacts. To address this problem, we recently proposed an enhanced algorithm for linear-array PAM by introducing dual apodization with the cross-correlation (DAX) method into TEA. But due to that the delay and sum beamformer used to create RX1 and RX2 is non-adaptive, the remaining X-type artifacts cannot be completely suppressed, yielding unsatisfactory image quality. This study aims to propose an improved version by combining DAX and robust Capon beamformer (DAX-RCB). Methods: Different from the delay and sum beamformer in the DAX-TEA algorithm, in the proposed version, the two sets of channel signals from a pair of complementary receive apodizations are beamformed by the RCB method, which may make passive cavitation images much less sensitive to X-type artifacts. The performance of the DAX-RCB algorithm is validated by simulations and in vitro experiments and compared with the initially used TEA algorithm and the previous DAX-TEA and RCB algorithms. Four indexes, including passive energy beam (PEB) size, image signal-to-background ratio (ISBR), energy estimation ratio (EER), and computing time, are used to evaluate the algorithm performance. Results: Consider an example of the 8–8 alternating pattern (a pair of complementary apodizations are obtained by extracting eight elements every eight elements), the experimental results show that the A_–6dB area (2D PEB size) of the proposed DAX-RCB is significantly reduced by 11.0 and 6.8 mm² when compared with TEA and DAX-TEA and is not significantly reduced when compared with RCB, the ISBR is significantly improved by 19.6, 10.8, and 5.6 dB compared with TEA, DAX-TEA, and RCB, and the EER of DAX-RCB is over 90%. The simulation tests indicate that the DAX-RCB algorithm is also applicable to the image enhancement in the double-source scenario and the high-level noise scenario but at a risk of low energy estimation. The improvement of algorithm performance is accompanied by an increase in the computing time. The proposed DAX-RCB consumes 113.3%, 29.5%, and 17.8% more time than TEA, DAX-TEA, and RCB. Conclusions: The proposed DAX-RCB can be considered as an effective reconstruction algorithm for passive cavitation mapping and provide an appropriate monitoring means for ultrasound therapy, especially for cavitation-mediated applications.